Date: 2026-10-11 (Sunday, 1:30 PM EDT) Attendees: Michael Houck (michael@ryevalley.co), Emil Barr (emil@flashpass.school), David (DataCow, david@datacow.co) Granola: https://notes.granola.ai/d/4295ceab-0416-401d-ba37-2b670b7f5a44
Context: GovGrid pilot buyer conversation, part of the Michael + Emil data licensing push (see Houck / Emil Weekly - Oct 9, 2026). Michael booked via DataCow’s site form; Emil did most of the pitching. This is Emil’s engagement, not a Rye Valley-owned business decision.
Summary
- DataCow (San Francisco) licenses company data for AI model training. Their process: high-level scoping call, mutual NDA, connect data via API, automated eval script, price quote (a few days, up to 1-2 weeks). They work 7 days a week.
- GovGrid pitch (Emil): government contracting business active in ~15 states with master contracts through state attorneys general covering all agencies. Can pull non-PII state data: agency emails (internal and inbound), SOPs, training manuals, new-hire guides, internal presentations, and back-office finance data (treasury, auditor, budget accounting).
- Scale and quality: terabytes to tens of terabytes, hundreds of millions of data points. ~80% retrieval rate on requests. ~5-10% arrives partially redacted by the state (names, occasionally whole emails); GovGrid does no redaction of its own. Documents are standardized within a state but meaningfully different state to state.
- Exclusions: PII, anything framed as national security (e.g. state National Guard). Teams/chat logs are rare; state agencies run almost entirely on email.
- Moat claim: no other known multi-state contractor selling state data to AI labs; state relationships take years and 2-3 year sales cycles to build.
- Exclusivity: open to it, but only at scale and only after a working pilot.
- Optional cleanup: in-house engineers can dedupe, cull redacted items, and standardize, priced higher than a raw dump.
- DataCow interest: David flagged back-office finance/accounting data as a current area of interest. This deal is outside his usual direct-business-data lane; he will bring in a colleague better suited to pass-through government data.
- Note: on this call the state-government source was disclosed up front, unlike the “lead vague” outreach framing agreed Oct 9 (decision). DataCow had scheduled the call off the site form, so the context was already a live conversation.
Decisions
- GovGrid deal posture stated to DataCow: start with a small scoped pilot, no exclusivity commitment until the working relationship is proven. See decision page.
Action items
- David (DataCow): send mutual NDA shortly after the call
- David (DataCow): loop in colleague who handles government pass-through data; aim to make an offer quickly
- Emil / Michael: define the “dream pilot” scope with DataCow, likely back-office finance/accounting data
Notable transcript excerpts
Emil on the data access mechanism:
“What we’ve often done with these states is we actually have contracts with the state attorney general. That then applies to any agencies. So we could theoretically go in and say, hey, we want all of the emails pertaining to X, or we want all of the financial data last year from every agency.”
Emil on the moat:
“What’s made our business unique and built us a moat is that every state government sucks… you can’t call the governor of California and say, hey, we want to buy all your government’s data.”
Emil on exclusivity and pilot:
“I don’t want to commit to exclusivity with any one… broker sort of like vendor, until we’ve actually had a chance to work with them… if we’re then going to turn around and do a 100x bigger deal over like the next 6 months, we want to make sure that that’s a good partner.”
Emil on finance data:
“These state budgets, like in Ohio, it’s in the hundreds of billions of dollars… there is so much financial insight in like these hundreds of billions of dollars being spent that no one else is looking at.”
David on next steps:
“This is a little out of my wheelhouse right now, although I do have a colleague who I think may be better for this and would be able to like navigate the situation better. And ideally we can, you know, make you an offer and move quickly.”
Full transcript in Granola: https://notes.granola.ai/d/4295ceab-0416-401d-ba37-2b670b7f5a44